Time frame on next gen LED?

I really like my SF L4...but I want more power! Muahahahaha! Anyone know when the next generation of LED will be out on market? I think my L4 is the 5 Watt Luxeon. I heard that 10 Watt LED will be here soon? Any ideas?

There have been rumors about 20W parts from Lumileds circulating about at industry conferences and trade shows. There's a publicly-available presentation from Lumileds that shows a mysterious unit on a large heatsink.

When is something that noone knows. Apparently, Lumileds feels no pressure to release new products from other manufacturers (many thought that Cree's XLamp would prod them into action, but the XLamp doesn't seem to be widely available yet).

I expect that Lumileds will increase the efficiency of their products before any fundamentally new products are released. Everyone really just wants more light, not more power.

I have a feeling the efficiency will have to be increased before more powerful LEDs are offered due to heat problems with the larger versions may be akin to the heat on computer CPU chips... first it was no heat sink, then fins, then larger fins, then small fans, bigger fans, and now almost need water cooling/refridgeration. That is not a good thing for a light to have to use because the drives the efficiency way down due to cooling costs.

I agree about the efficiency. I think it's more important to have more efficient LEDs than higher powered ones. You double the efficiency and you double the output. That is certainly better than doubling the output by doubling the power (and hence the waste heat). Long term LEDs need to get more efficient anyway if they are to compete with other forms of lighting.

I think 2005 should be very interesting in that things which were being worked on in the labs for a long time will finally be put into production. I think before the year is out 50 to 70 lm/W white LEDs will be the norm, both in 5mm and high power, with 100 lm/W perhaps year or two after that.

I sincerely hope not.
Or I will have to buy a whole load of new lights before having enjoyed the actual ones adequately ... /ubbthreads/images/graemlins/icon15.gif
bernhard

I hope there will be perhaps 10-15% increase in efficiency by the end of next year but I wouldn't hold my breath on anything over 40 lm/W available to the general public. Even if they can get efficiency up to 50lm/W is that at full output or idling at 100-200ma on luxeons? I have a feeling breakthroughs will be gradual just like in the processor business. I sincerely hope I am proven wrong.

[ QUOTE ]
Kiessling said:
I sincerely hope not.
Or I will have to buy a whole load of new lights before having enjoyed the actual ones adequately ... /ubbthreads/images/graemlins/icon15.gif
bernhard

[/ QUOTE ]
That's the nice thing about modular lights. Unless battery formfactors change, they'll be good for years. /ubbthreads/images/graemlins/smile.gif

(Bernie, is that a Replicator you are using as your avatar? If so, please be veeery careful with it. /ubbthreads/images/graemlins/icon15.gif )

Larry

Actually, until the last year or two processor speed was increasing at about 60% per year and hard drive size was increasing at 100% per annum. Unfortunately, both are now reaching hard physical limits where increases will come about much more slowly. It seems like the fastest processor has been around 3.6 GHz forever, and hard drives were stuck on 80 GB/platter for well over a year until the recent 100 GB and 133 GB per platter models.

LEDs won't start reaching similar limits until perhaps 150 lm/W. After that, it may take a decade to reach 200 lm/W. I'm predicting dramatic increases for the simple reason that production white LED efficiency hasn't gone up by much in the last year or two (perhaps 30%) yet we have had laboratory samples with more than twice the efficiency for quite some time now. It's been well over a year since Cree announced a 74 lm/W efficiency of 5mm LEDs in special laboratory packages (and 65 lm/W in standard ones). I believe this was with a blue chip with 34% wall-plug efficiency. Cree now has production blue chips with 37.5% efficiency, and 42% efficiency ones are in the testing stages. This means that with special packaging a 90 lm/W white LED is possible right now as a prototype. I realize it takes time for these improvements to reach production but no more than maybe two years, if that. Nichia for one announced that its 60 lm/W 5mm LEDs would go into production in 2005. This is why I feel next year should bring about dramatic increases. We may have the manufacturers playing oneupsmanship trying to outdo each other in efficiency by rolling their lastest prototypes into production faster than normal. I'll be very, very disappointed if all we have by the end of next year is 35 to 40 lm/W. This means the manufacturers are trying to milk the market by holding back on the latest technology for as long as possible. Given that somebody is bound to roll out their latest and greatest, if for no other reason than bragging rights, I don't believe that will happen.

[OT] Larry ... my avatar is a KGC-000 King Crab BattleMech from the Battletech tactical boardgame ... and in fact a photo of a Bernhard-painted 25mm figurine of the thing ... /ubbthreads/images/graemlins/smile.gif [/OT]
bernhard

Whoa...good stuff here. Thanks guys.

Prototypes from companies that don't seem to have much of a market share don't impress me. I know nichia has made a 2watt LED but can you get them at a competing price between 1 and 3 watt luxeons? Nichia's 5mm LEDs are too expensive and not a value and lights with them in them only have 1 usually vs a bunch of chinese clones for less cost. Unless nichia and cree change their market strategies I figure we have to hope lumileds and the chinese make the breakthroughs because they are the ones improving stuff and selling it.

Actually, I think a lot of the Chinese LEDs are using either Cree or Toyoda-Gosei dies. And they do seem to improve on a regular basis. Around January when I purchased my first LEDs from Chi-Wing I noticed that the best Chinese LEDs on eBay were around 12,000 mcd with either a 15° or 20° viewing angle. Now I've seen ones for sale as high as 22,000 mcd, also with the same viewing angle. All figures are of course at 20 mA. A 22,000 mcd, 20° LED puts out roughly 2.5 lumens (2.2 in the main beam and another 10% to 20% in sidespill). At the typical Vf of around 3.2V this puts efficiency right around the 40 lm/W mark (compared to the low 20s about a year ago). If 2005 has a similar efficiency increase then we should see 60 to 70 lm/W by the year's end.

As for Nichia, I'll agree that they're overpriced and also hard to get. Even the ones allegedly in production since this summer which are supposed to average 40 lm/W are nowhere to be found. Every time I see Nichia's for sale, it's the same 9200 mcd, 20° part which has been around for at least the last two years. Apparently it takes a really long time before the better stuff filters down to their distributers.

A 20 watt LED light would be great. Too bad it would have to be used ONLY under water to stay cool. /ubbthreads/images/graemlins/tongue.gif

The 22,000mcd ones either are rare, have a narrower viewing angle or twice as much as a 1/2 output 10,000mcd one. I think it would take about 80-100lm/W in 5mm before they become useful as single LED lights compared to todays luxeons. If the 22,000mcd ones are legit I would be tempted to buy one to replace the one in my UK2AAA to double the output of it.

How about 575 lumen / Watt at 535 nm?
Check out US patent #6686691 by Lumileds on
http://www.freepatentsonline.com

C6H6

C6H6:

Actually, the 575 lm/W refers to the efficacy of the emitted radiation, not to the efficiency of the device itself (which would need to be 100% to match this value). The blue plus green phosphor approach probably narrows the bandwidth of the emitted light so that shorter wavelengths to which the eye is less sensitive are not present.

Lync_Arc:

Here is one eBay auction for 22,000 mcd and 20° to 25° viewing angle. They are a little pricey at $0.74 each shipped which indicates to me that they are probably from a select bin. If they aren't what is claimed, the seller would likely receive negative feedback.

There have been LEDs in the past with 20,000+ mcd intensities but these had very narrow viewing angles (5° to 10°). Assuming all the specs here are legit, these LEDs should output anywhere from about 2.5 to perhaps 3 lumens. This would mean efficiencies on the order of 40 lm/W or better. This is certainly believeable since the prime bins from Lumileds also have similar efficiencies.

100 lm/W white LEDs would be quite useful when they appear. For starters, since there will be less waste heat this would mean that they can be speced at 30 mA instead of 20 mA. Given typical Vf values running at 30 mA equals about 0.1W input power, which translates to an output of roughly 10 lumens. I could have a very nice bike light with maybe 20 of these. Even one would be quite useful.

Yes, but depending on how much luxeons increase people may desire less LEDs putting out more of a spot than a lot putting out a flood. The idea is to get down to just a few LEDs instead of a huge array of them. As for the 22,000 20 degree LEDs I wonder if anyone on CPF has ordered any. If not it could be the people that buy them don't know the difference between them they just look brighter than the 10,000mcd ones they bought before.

Ahhh, I forgot about that. /ubbthreads/images/graemlins/blush.gif
Well at least I can look for the possibility of have flashlights with multiple luxeon leds on it. I think I saw one, but it was on a big maglight. I rather have one the size of a surefire.

Collimation of multiple light sources is a real challenge.

I've played with several array lights and can say with confidence that none throw as well as the average single-source light with a modest reflector.

Take something like a Trilight 3 - it features 3 Luxeon IIIs each with a 17mm or 21mm reflector, but it won't throw anywhere near as far as a single luxeon in a 55mm Mag reflector occupying the same head assembly.

Parallelism produces more light, but at the cost of throw and requires some tweaking to get beam convergence.

Unlike 5mm LEDs, Luxeons require optics to produce a beam. Without optics, a high-dome Luxeon would make for a decent hemispherical area light (behind a diffuser)...

Here's a graphic I drew to try to illustrate the point. The single-source light on top has a 3-degree beam angle. The multi-source light below has 3x 15-degree beam angles. Notice that more "flux lines" strike the target (to the right of the lights - same "height" as the "flashlights") and are more dense than the triple-source target?

Another problem with multiple LEDs is it makes for more cost. A single LED costs about as much as a incan light bulb to make I figure perhaps a little less although bulbs are selling for a lot in the stores I figure they are priced high because few people replace bulbs often. I figure when they get LEDs that put out as much as a typical 2D flashlight that costs $1.00 and can be put in a cheap plastic light with minimal heatsinking, and throw similarly the incan may have seen its last gasp for low end lighting.
I think one breakthrough that may not happen but would help tremendously is lower voltage white LEDs. If they could make them run at 2.4 volts or 1.2 volts at rated output then all the excess batteries and boost/buck circuitry may not be as much required. OR.. high voltage LEDs... that would lower the current needed to drive them dramatically. Imagine a 12v LED that takes 100ma of current for 1.2watts, this would be an interesting situation for sure.